Hand held data collector and analyzer systems are typically used to collect vibration data from machines for use in predicting maintenance requirements. For example, a typical data collector and analyzer instrument may be programmed to receive a route from a host computer, and such route would include a list of machines, test points, and a set-up condition for each test point. There may be thirty machines in the route with ten test points on each machine, and for each test point there may be specified a frequency range to be analyzed, a type of analysis to be performed, a particular type or set of data to be stored, and similar other parameters. In response to commands from the user, the hand held instrument prompts the user as to the identity of the machine and the test point to be monitored, and it automatically sets up the instrument, for example, to accept the specified frequency range for the test point, perform the specified analysis and store the specified type or set of data. A Fast Fourier Transform analysis may be performed on a pre-selected frequency range of the data and all or part of the resulting frequency spectrum may be stored and displayed. As the user progresses through the thirty machines and the corresponding 300 test points, he collects and stores vibration data which is subsequently transferred to the host computer for long term storage and further analysis.
Such hand held instruments are subject to rather severe weight and size constraints imposed by the need to be hand held and also must operate in a very hostile environment, such as a power plant or a heavy industrial manufacturing facility. In this environment, the instruments are exposed to a wide variety of vibration as well as a wide variety of temperatures, pressures, gas, dust and other atmospheric conditions. In such environment, the hand held instruments must be able to monitor vibration accurately in the presence of extreme noise. For example, one may wish to monitor a particular test point for a possible crack in a shaft bearing, where the bearing is rotating at 200 hertz, or 12,000 RPM, for example, and the bearing has twenty ball bearings. A crack in such bearing would produce a relatively high frequency, low amplitude, vibration or click each time a ball bearing passed over the crack. If such bearing is located in a typical factory, it will be operating in the presence of noise vibration from many sources such as stamps, presses, roller mills, conveyors, pumps, motors, etc.
Thus, to distinguish a click associated with a crack in a bearing, careful analysis of the frequency content of the monitor vibration signal is performed. Preferably, a user would select a frequency range of interest and look for a specific frequency that indicates or may indicate a cracked bearing. Typical known data collector and analyzer systems use analog filters to remove some of the unwanted frequency components. Usually a frequency spectrum is generated in a specific range and then specific data from the spectrum is inspected, compared or just stored.
One approach to providing the frequency analysis capability has been to provide a variable low-pass analog filter whose cutoff frequency can be controlled, and the output of the filter is sampled to provide a digital signal that is operated on by a computer to produce a frequency spectrum. This approach is generally acceptable, but it is not without limitations.
For example, analog low-pass filters will produce a certain amount of distortion of amplitude, frequency and phase, and variable low-pass filters tend to have greater distortion than fixed low-pass filters, particularly at or around the upper cutoff frequency of the filter. Such distortion is acceptable for many applications, but it certainly is not a desirable characteristic. In certain sensitive applications, such distortion may disqualify an instrument.
In addition to limitations at the higher end of the frequency range, some prior art instruments are known to be inaccurate and even unstable at low frequencies, particularly at frequencies on the order of one hertz or less. This characteristic usually results from circuit designs focused on higher frequencies because such frequencies are generally of greater interest. Again, however, in some applications accurate monitoring of low frequency signals is critical.
Another frequency related problem is the manner in which hand held data collector and analyzer instruments produce output that is limited to a frequency band of interest. One approach might be to use variable filters, such as switched capacitor filters, for which both a low-pass and a high-pass cutoff may be specified, or banks of filters may be used. Generally, this approach is not practical for a hand held device.
Another approach has been to provide a variable low-pass filter whose output is converted to a digital signal, and a Fast Fourier Transform is performed on the digital signal to produce a spectrum. Then, only the frequency band of interest in the spectrum is displayed or stored. This technique can be wasteful in terms of time, processor demands and accuracy. For example, to provide a 100 line spectrum between 9 kilohertz and 10 kilohertz, such an approach calculates a 1,000 line spectrum between 0 and 10 kilohertz, essentially discards the lower 900 lines, and displays the upper 100 lines. The discarded 900 lines represent waste.
These and other problems are addressed by the present invention which uses a fixed filter design to overcome the problems associated with variable filter designs. The computer source code for the present invention is listed in its entirety in the appendix. In accordance with the present invention, a hand held vibration data collector and analyzer system includes a vibration transducer producing an analog vibration signal. An input signal conditioning circuit receives and conditions the analog vibration signal from the vibration transducer to produce a conditioned analog signal, and it also includes a fixed frequency low-pass, or anti-aliasing, filter having an upper cutoff frequency set at a desired frequency for producing the conditioned analog signal with a desired frequency range. The input signal conditioning circuit further includes amplifiers for producing the conditioned analog signal at a desired amplitude, a D.C. offset circuit for removing D.C. components from the analog signal and an analog to digital converter for receiving and sampling the conditioned analog signal to produce a digital signal which is transmitted to a data processor for producing desired digital data.
The processor includes a transformer for selectively operating on the digital signal, such as performing a Fast Fourier Transform, and producing a frequency spectrum from the digital signal. The data processor also includes a selector for selecting and producing select data for storage from the digital signal or the frequency spectrum. A keyboard is interfaced with the data processor for inputting commands and data, and a display is interfaced with the data processor for displaying information. A memory, interfaced with the data processor, stores information including the select data and means are provided for transferring information stored in the memory to another computer.
One advantage of the system described above is that the fixed filter does not produce the same amount of distortion that is normally associated with variable low-pass filters, especially in the vicinity of the upper cutoff frequency of the filters. In addition, a selected frequency range of interest will normally be distant from the cutoff frequency of the fixed frequency low-pass filter, which is not the case in designs using variable frequency low-pass filters, as discussed above. Thus, with the design in the present invention, the distortion in the vicinity of the cutoff frequency of the fixed frequency low-pass filter will normally not have an impact on the operation of the instrument.
In a preferred embodiment, the input conditioning circuit includes hardware for receiving and sampling the conditioned analog signal at a sample frequency that is substantially greater than the maximum frequency of interest to produce a digital signal and further, for digitally low-pass filtering and digitally decimating the digital signal to produce the digital signal having a reduced sample rate and a selected upper cutoff frequency. In this embodiment, the data processor also includes a digital filter and decimator for optionally and selectively reducing the sample rate and frequency content of the digital signal to produce a modified digital signal. Thus, in this particular embodiment, two separate portions of the system are designed to perform digital low-pass filtering and digital decimation.
In general, hardware performing digital filtering and decimation is fast and precise, but generally, the variability of the output sample rates is limited. Digital low-pass filtering and decimation in the data processor is typically slower, but one may precisely choose the output sample rate and even the level of precision desired. By dividing the duty of digital filtering and decimation between hardware in the input signal conditioning circuit and the data processor, one may minimize the demands on the data processor without sacrificing precision, speed or variability. Also, by dividing this digital function between two portions of the system, one may efficiently impose other operations between the two digital filtering and decimation operations, again, without placing an undue or unrealistic demand on the data processor.
For example, in a preferred embodiment so-called true zoom processing is facilitated by the two digital filtering and decimation operations. In this embodiment, the data processor again includes a digital filter and decimator as described above. The data processor further includes a zoom processor for selectively and optionally operating on the digital signal from the conditioning circuit. To begin the zoom processing, the hardware of the conditioning circuit is actuated to digitally filter and decimate the input digital signal to produce a digital signal at a sample rate and with a frequency content as low as possible, but still greater than an upper frequency, which is the maximum frequency of interest in a frequency band of interest. The zoom processor operates on the digital signal by frequency shifting the digital signal and at least low-pass filtering the frequency shifted digital signal to produce a zoom digital signal. The zoom digital signal is a selected band of frequencies in the digital signal ranging from a selected upper frequency, the maximum frequency of interest, to a selected lower frequency, and having a center frequency. Preferably, the zoom processor multiplies the digital signal by a function equal to e.sup.2.pi.i(F0)/(Fs), where F0 equals the center frequency of the selected band and Fs equals the sample rate of the digital signal.
This multiplication will effect a frequency shift of the digital signal such that the center frequency is shifted to zero. Then, preferably, the frequency shifted signal is subjected to low-pass filtering where the upper cutoff frequency of the low-pass filter is equal to: (.function..sub.u -.function..sub.b)/2 where .function..sub.u =the upper frequency and .function..sub.b =the bottom or lower frequency in the frequency band of interest. A Fast Fourier Transform operation is performed on the low-pass filtered, decimated, and frequency shifted signal to produce a frequency spectrum that is then shifted up by f.sub.b, the lower frequency, to become a frequency spectrum from f.sub.b to f.sub.u. By shifting the frequency of the digital signal down before performing the filtering and transform operations, the speed and efficiency of the zoom processing is dramatically increased without any loss of accuracy.
To accommodate and process high frequency signals produced by the preferred embodiment of the present invention, the digital signal processor includes a digital signal processor (DSP) connected serially to receive the digital signal from the conditioning circuit for operating on the digital signal and, at least, for independently performing Fast Fourier Transforms to produce a frequency spectrum from the digital signal. The data processor further includes a central processing unit for controlling the operation of the system, including the operation of the DSP. In addition, a direct memory access (DMA) is provided in the CPU for transferring data directly to memory from the DSP to memory without interrupting the CPU.
In an alternate embodiment, the system includes a main conditioning circuit and an optional conditioning circuit, both of which include fixed frequency anti-aliasing filters, amplifiers and analog to digital converters. The main conditioning circuit includes a main microprocessor controller for controlling the main conditioning circuit, including the main analog to digital converter. The optional conditioning circuit does not include a separate microprocessor controller. Instead, control over the optional conditioning circuit is accomplished by and through an optional DSP.
Thus, this embodiment includes a main DSP connected to serially receive the main digital signal and an optional DSP connected to control said optional conditioning circuit and connected to serially receive the digital signal from the optional conditioning circuit. A CPU controls the overall operation of the system including issuing commands to the main and optional DSP's and the main microprocessor controller for controlling the optional conditioning circuit and the main conditioning circuit. In the preferred embodiment, the optional conditioning circuit is a separate optional circuit board and includes a plug system for connecting and disconnecting the optional circuit board to and from the system.
In the preferred embodiment, the memory for the system includes a main bank of random access memory (RAM) and a memory card that is selectively plugged into and out of communication with the data processor. If a memory card is present and functioning properly, the data processor can store data directly to the memory card. However, if the memory card is missing, the data processor can store data in a pseudo-card that is configured in the RAM. In the preferred embodiment the user selects either the installed memory card, or the pseudo-card in RAM as the storage media. A pseudo-card in RAM is created by lower level programming and is designed to configure the memory in RAM in the same manner as a memory card. Thus, the upper level programming in the data processor stores data in only one format, namely the memory card format. To the data processor, the pseudo-card in RAM appears to be identical in data format to the actual memory card.
As previously mentioned, the problems of the prior art relating to frequency are not limited to high frequency problems. Inaccuracy and possible instability also appear in the very low frequencies. In the preferred embodiment, the input conditioning circuitry includes first and second analog to digital converters. The first analog to digital converter receives and samples the conditioned analog signal at a relatively high sample rate to produce a first digital signal, and the second analog to digital converter receives and samples the conditioned analog signal at a relatively low sample rate to produce a second digital signal. The second analog to digital converter includes precision quantizers that are accurate and stable in the conversion of low frequency signals in the range of about one hertz. A data processor receives and processes the first and second digital signals to produce select data for storage.